LT's automatic lecture subtitling for German launches university wide at Lecture2Go
18 October 2021, by Benjamin Milde
The university-wide media plaftform Lecture2Go has integrated LT's automatic lecture subtitling solution for German [1] into their production system! This feature enables any lecturer from Universität Hamburg to automatically subtitle any uploaded media files to the Lecture2Go platform.
LT members started an effort to make Open Source ASR models for German available to the community in 2015 [2,3]. The training scripts and pre-trained models have matured over the years and the project has grown immensly in popularity. It can now be used as a freely available building block for German ASR applications with production-ready recognition accuracy!
We continue to demonstrate how foundational research can lead to usable applications. Congratulations to all former and current LT members that were involved in this project!
[1] Milde, B., Geislinger, R., Lindt, I., Baumann, T. (2021): Open Source Automatic Lecture Subtitling. Proceedings of ESSV 2021, pages 128-135, virtual Berlin. (pdf, code)
[2] Radeck-Arneth, S., Milde, B., Lange, A., Gouvea, E., Radomski, S., Mühlhäuser, M., Biemann, C. (2015): Open-Source German Distant Speech Recognition: Corpus and Acoustic Model. In: Text, Speech and Dialogue. Proceedings of the 18th International Conference TSD2015, pages 480-488, Plzen, Czech Republic (pdf)
[3] Milde, B. and Köhn, A. (2018). Open Source Automatic Speech Recognition for German. Proceedings of 13th ITG Conference on Speech Communication 2018, pages 251-255, Oldenburg, Germany. (pdf, code)